RSNA Screening Mammography Breast Cancer Detection Dataset from Two Clinical Sites

Just under 20,000 screening mammography imaging studies were collected by the Radiological Society of North America (RSNA) from two clinical sites. The dataset is intended to support the development of machine learning models for early breast cancer detection. According to the description, breast cancer is the most commonly occurring cancer worldwide, with 2.3 million new diagnoses and 685,000 deaths in 2020.

Use Cases

• Training computer vision models for breast cancer detection based on screening mammograms.
• Developing AI tools to assist radiologists in evaluating mammography studies.
• Researching methods to reduce the cost and expertise required for large-scale screening programs.
• Benchmarking model performance on a clinical dataset of just under 20,000 imaging studies.

Strengths

• Contains just under 20,000 imaging studies, providing a substantial base for model training.
• Data was collected by the Radiological Society of North America (RSNA), a professional medical society.
• Focuses on screening mammography, a critical tool for early cancer detection where mortality has dropped by 40% in high-income countries since the 1980s.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for individual images or associated metadata is unknown, which may limit suitability assessment.
• Data may reflect geographic or institutional bias inherent to the two collection sites.

Provenance

Source

Radiological Society of North America (RSNA)

Collection Method

Collected screening mammograms and supporting information from two clinical sites.

Time Range

null

Freshness

Last update date is unknown; freshness unverified.

Geography

null